******************* FILE 1 of 6 FILES ********************** (C) 31 JUL 88 Eric Gustafson, N7CL 2018 S. Avenida Planeta Tucson, AZ 85710 DPLL Derived Data Carrier Detect (DCD) For Filter Based and Single Chip Modems INTRODUCTION If you have a TNC which uses either the AMD7910 or the TCM3105 single chip modem, or a TNC which uses a modem based on audio filters like the PK-232, you can vastly improve the DCD performance of your modem for packet radio use. These single chip modems were originally designed for land line use. The designers, who had no idea that the chips might one day be applied to a radio system, made some assumptions about the incoming signal that simply do not apply to the radio environment. The data carrier detect function for them was not nearly so critical a function of the modem as it is for us on a busy packet radio channel. For the intended purpose of these chips, there was expected to be only 2 stations involved on any 1 channel at 1 time and these stations were connected by a nice quiet twisted pair. Under these circumstances, the Carrier Detect (CD) function built into these chips is entirely adequate. In the packet radio environment, the built in CD function is next to useless. Since I can't make the same defense for the designers of filter based modems specifically intended for packet radio from the beginning, I won't try to speculate about what drove their design decisions. The circuit presented here will allow your TNC to be used with unsquelched audio thus avoiding the unnecessary delay of the squelch circuit found in typical VHF FM radios. This circuit also provides several other important beneficial characteristics for the DCD system. First, since the assumptions used when the TNC software was written depend on DCD representing the presence or absence of a data carrier on the channel, it is important that the DCD circuit be able to distinguish a data carrier from noise or other non packet signals to a reasonable degree. The DCD circuits which simply detect the presence of ANY type of signal or noise on the channel are simply inadequate to this task. Since the DCD circuit presented here is based on the update signals in a Digital Phase Locked Loop (DPLL) which recovers both baud clock and data from an NRZI packet data stream, its output represents true detection of the data carrier. Second, once a data carrier decision has been correctly made, it is important that the DCD indication remain valid through short fades, collisions, and while a signal too marginal to decode is on the channel. This is accomplished by providing a DCD "hang time" of approximately 5 to 8 character periods (this can be optimized) to hold the DCD output true through short dropouts from the above causes. This prevents a queued up TNC from piling on collisions, transmitting over a station which has a marginal signal, and beginning to transmit over a station which is still transmitting but whose signal received a short multipath hit during the packet. Third, it is important that the DCD system NOT be sensitive to audio amplitude variations. It should respond in exactly the same way for any signal that the modem is capable of decoding regardless of absolute input amplitude. Since this DCD circuit operates from the data recovered by the modem, all amplitude information is suppressed before the DCD circuit even sees the signal. NOTE! If your TNC uses the EXAR 2211 demodulator, this new circuit is unnecessary for you. Your existing DCD circuit can be more easily modified for correct operation without this circuit. The modification procedure for the 2211 demodulator is presented elsewhere and not repeated here. (C) 31 JUL 88 Eric Gustafson, N7CL 2018 S. Avenida Planeta Tucson, AZ 85710 DPLL Derived Data Carrier Detect (DCD) For Filter Based and Single Chip Modems CIRCUIT DESCRIPTION The circuit diagram is presented in Figure 1. This is an ASCII representation of the schematic diagram. While this isn't really a proper "standard" diagram, I believe it is readable enough to be used to duplicate the circuit. It has the beneficial characteristic that it requires no CAD or special graphics software to be able to view the diagram. Thanks to Mykle Raymond, N7JZT, for making up this BBS forwardable ASCII schematic. Figure 1 is contained in the last file in this series. The circuit consists of the state machine used in the TNC-2 and some delay elements used to make the DCD decision. The state machine is formed from the 74HC374 and the 27C64 chips. The 74HC14 is used as a pair of retriggerable delay elements and for signal inversion and buffering. The 27C64 with the state machine code already burned into it can be obtained directly from TAPR. If you wish to use this source for the part, please call Chris at (602)-323-1710 for price and availability information. This same code is in the state machine ROM in any full TNC-2 clone which uses the 2211 demodulator and Z80 SIO. If sufficient interest is shown in this circuit, maybe we can cajole TAPR into making circuit boards available. This would vastly reduce the wiring task. One of the state machine signals (which was not used in the TNC-2) appears on pin 19 of the 27C64. This signal is the DPLL update pulse. As long as the DPLL is correctly locked to the incoming data, no pulses will appear on this pin. When the DPLL is not locked to an incoming data stream, there will be a continuous stream of pulses on this pin. The DPLL update signal is used in this circuit to retrigger the first delay element so that it never times out so long as DPLL update pulses are present. If the pulees disappear, the delay element times out and generates the DCD signal. The output from the first delay element keeps the second delay element triggered so long as DCD is true. When DCD goes false, the second delay element begins a timeout sequence which keeps the DCD output true until the timeout period expires. This is the source of the DCD "hang time". While the circuit presented here is primarily intended for 1200 baud VHF FM operation, it will also work well for 300 baud HF packet work. If this is your application, the time constants on the delay elements will have to be adjusted. The time constant of the "hang" generator (0.47 uF cap) will have to be increased for 300 baud operation so that the total capacitance is 2.0 uF. The time constant which is optimum for the DCD generator (the 0.1 uF cap in fig. 1) will depend on a number of factors including the bandwidth of the radio used ahead of the modem. You should pick a value for the DCD generator delay capacitor such that the DCD circuit produces approximately a 10 percent duty cycle of false DCD "ON" time. The false DCD ON time should be observed while monitoring receiver noise on a channel which is ABSOLUTELY free of ANY narrowband signals which fall within the demodulator's passband. This includes CW, R????? internal receiver birdies, AM carriers, computer spurs, packet data carriers, etc. A good way to assure this is to let the receiver monitor the S-9 or greater output of a noise bridge with no antenna connected. Re??????????????????????????? itor will probably need to be somewhere in the range of 2 to 4 times the 0.1 uF value used for 1200 baud. Both negative true and positive true DCD outputs are provided so that you may use the polarity which is required by your TNC. Also, JMP1 and JMP2 allow the DCD circuit to be configured to operate correctly from either a positive or negative true CD output from whichever modem chip is found in your TNC. (continued in CIRCUIT_3 EOF Msg # 12779 Type:B Stat:$ To: ALL @OKIPN From: N7CL Date: 28-Aug/2323 Subject: Improved DCD circuit_3 Bulletin ID: 6171_W1FJI Path: AD8I!WB8ICL!N8GTC!KI4UN!KF4NB!WB9TPG!K3VYY!K4EID!K4EID!KE7CZ!W1FJI ******************* FILE 3 of 6 FILES ********************** (C) 31 JUL 88 Eric Gustafson, N7CL 2018 S. Avenida Planeta Tucson, AZ 85710 DPLL Derived Data Carrier Detect (DCD) For Filter Based and Single Chip Modems TNC SIGNALS Once you have constructed the DCD circuit, you will have to obtain some signals from your TNC for the new DCD circuit to use. You will alircuit to be substituted for the normal DCD signal used in the TNC. The signals required for the DCD circuit operation are: 1. A sample of the data recovered by the demodulator in the modem chip. 2. A sample of a clock which has a frequency of either 16 or 32 times the baud rate (X16 or X32 baud clock). 3. The intercepted Carrier Detect (CD) signal from the modem chip. This is the CD generated by the modem chip based on amplitude of the input audio. 4. A source of + 5 volts. If you use all CMOS parts, the current requirements are minimal. The 74HC14 MUST be a CMOS part for the circuit to work properly. 5. Ground There are so many different TNCs to which this circuit can be applied that I cannot give specific interface information for all of them. However, I can provide signal pin numbers for the 2 land line modem chips most frequently encountered and I can help with signal locations in the AEA PK-232 and PK-87, the Kantronics KAM, and the Pac Comm TINY-2 TNCs. The signals of interest on the AMD7910 modem chip are: 1. Receive Data output (RD)-----> pin 24 2. Carrier Detect (CD)----------> pin 25 This signal is negative true for the 7910 chip. The signals of interest on the TCM3105 modem chip are: 1. Receive Data output (RXD)----> pin 8 2. Carrier Detect (CDT)---------> pin 3 This signal is positive true for the 3105 chip. 3. In TNCs which use the TCM3105 chip but do not provide another source of the baud clock, like the Kantronics KAM, you can use the signal at pin 2 of this chip. This signal is very close to 16 times the baud rate (19.11 KHz instead of 19.2 KHz for 1200 baud). TNC INTERFACE If your TNC has provision for a TAPR style modem disconnect header, these signals (including the X16 or X32 baud clock) will be easily located and conveniently interfaced at this header. If it doesn't have this header, you will have to fish around in the circuit of your TNC on your own to locate them. SHAME ON THE MANUFACTURER OF A TNC WITH NO MODEM DISCONNECT HEADER!! The absence of a standard modem disconnect header means you may not CONVENIENTLY use ANY external modem with the deficient TNC. Using a standard disconnect system, the external modem can provide a front panel switch to allow you to select between the external and the internal modem. Modems which you might like to interface without loosing the use of the internal AFSK modem would include the BPSK / MANCHESTER FM modems required for several of the satellites. In any case, the DCD signal currently used in your TNC will have to be disconnected and rerouted through the new circuit. STANDARD HEADER SIGNALS The signal locations on the standard modem disconnect header are as follows: Receive Data is obtained from header pin 18. Carrier Detect is obtained from header pin 2. DataCarrier Detect (DCD) is inserted at header pin 1. Jumper from header pin 1 to header pin 2 is removed. The X16 (TNC-2) or X32 (TNC-1 and possibly TNC-2 clones using an 8530 HDLC controller instead of the Z80SIO) baud clock is obtained from header pin 12. COMMERCIAL TNC SIGNAL LOCATIONS Here is the information you need to find the proper signals in several commercially available TNCs. This is not intended to be a complete list by any means. It merely represents the units which I have had available to apply this circuit to here locally. These are the only TNCs for which I have specific interface information at this time. AEA PK-87 It is relatively easy to interface this new DCD circuit to the PK-87 in spite of the fact that there is no standard modem disconnect header. This is because there is no requirement to switch back to the internal DCD circuit once the modification is installed. If this were an external special purpose modem, you would be forced to open the TNC case and move several jumpers whenever you wished to change the modem being used. However, for our purposes in this modification, the jumpers provide convenient, easily located places to obtain and inject signals. The Receive data signal is obtained from the center pin of JP4. The Carrier Detect signal is obtained from the end of JP5 which connects to the modem chip. The DCD output signal from the new circuit is inserted at the center pin of JP5. Use the NEGATIVE TRUE output. The jumper originally installed at JP5 is removed. The DCD indicator on the front panel will show the action of the new DCD circuit. The X32 baud clock signal is obtained from pin 13 of U20 (a 74LS393 divider). Don't be tempted to get this signal from the "clock" line on J4, the external modem connector, as this is a X1 clock. I see so many manufacturers sending only the X1 baud clock out to an auxiliary modem connector that I have to wonder if they simply don't realize that synchronous modems require a clock which is a multiple of the baud rate. Asynchronous modems can cheaply and easily divide the X16 clock to get X1 but it is hard for synchronous modems to derive a faster clock from the X1 signal. (continued in CIRCUIT_4 EOF Msg # 12780 Type:B Stat:$ To: ALL @OKIPN From: N7CL Date: 28-Aug/2325 Subject: Improved DCD circuit_4 Bulletin ID: 7938_K4EID Path: AD8I!WB8ICL!N8GTC!KI4UN!KF4NB!WB9TPG!K3VYY!K4EID 7829 BF 5099 ALL N7CL K3VYY 0823/0506 improved DCD circuit_4 R:880823/0255z @:K4EID-1 Knoxville TN #:9174 O:N7CL S:880823/0506z R:880820/0417 @:KE7CZ Dewey, Az #:7499 Z:86327 R:880819/1444 @:W1FJI Scottsdale Az #:6166 Z:85251 ******************* FILE 4 of 6 FILES ********************** (C) 31 JUL 88 Eric Gustafson, N7CL 2018 S. Avenida Planeta Tucson, AZ 85710 DPLL Derived Data Carrier Detect (DCD) For Filter Based and Single Chip Modems COMMERCIAL TNC SIGNAL LOCATIONS (continued) AEA PK-232 The PK-232 is also relatively easy to interface in spite of the fact that AEA failed to implement a standard modem disconnect header even in their flagship TNC. For some reason, on this box AEA decided rather than bring out the wrong clock out to the external modem connector, they would bring out no clock at all. The Receive Data signal is obtained from the center pin of JP4. The Carrier Detect signal is obtained from the end of JP6 which is NOT connected to pin 3 of the external modem connector. The X32 baud clock signal is obtained from pin 13 of U8 (also a 74LS393 divider). The DCD output from the new circuit is inserted at the center pin of JP6. Use the NEGATIVE TRUE output. The jumper originally installed at JP6 is removed. To use the new DCD circuit with a PK-232 on VHF FM 1200 baud: 1. Set the audio level from the radio so that the tuning indicator "spreads" fully even on the station with the lowest transmitted audio level on the channel. 2. The existing DCD threshold control should be set so that the existing DCD indicator LED on the front panel lights up whenever there is ANY signal or noise input to the TNC from the radio. Be sure that even the station with the lowest amount of audio on the channel lights this LED. This LED should extinguish when there is no audio input from the radio (dead carrier from repeater etc.). If you wish to observe the action of the DCD signal generated by the new circuit, attach a 1 K resistor in series with a LED to the LED output of the new DCD circuit. The anode of the LED should be connected via the resistor to +5 volts. The cathode of the LED should be connected to the LED output of the new DCD circuit. If you wish, this LED can be mounted on the front panel where it is visible. Use a high effeciency LED. Pac-Comm TINY-2 The Pac-Comm TINY-2 does include a modem disconnect header. It is labeled J5 on their schematic diagram. For this they get +1 attaboy. Unfortunately, Pac-Comm attached J5 pins 11 and 12 to the wrong part of the baud clock divider chain. These header pins should have been in series with pin 1 of U10. This error results in there being a X1 baud clock signal on these pins instead of the X16 baud clock that should be there. So, even though they did implement a modem disconnect header, you will have to obtain the X16 baud clock from elsewhere on the circuit board. For this they get -1 attaboy (at least they are breaking even). The X16 baud clock signal is obtained from U10 pin 1. Receive Data is obtained from J5 pin 17. Negative true Carrier Detect (CDT) is obtained from J5 pin 2. NOTE! This is an inverted version of the CD output from the TCM3105 chip itself. Since this is a negative true logic signal, JMP1 on the new DCD circuit will be used instead of JMP2 which would normally be used for a TCM3105. NEGATIVE TRUE DCD from the new circuit is applied to the TNC at J5 pin 1. Remove the connection between J5 pins 2 and 1. The existing DCD indicator LED will NOT show the action of the new circuit. If you wish to observe the action of the DCD signal generated by the new circuit, attach a 1 K resistor in series with a LED to the LED output of the new DCD circuit. The anode of the LED should be connected via the resistor to +5 volts. The cathode of the LED should be connected to the LED output of the new DCD circuit. If you wish, this LED can be mounted on the front panel where it is visible. Use a high effeciency LED and increase the value of the series resistor to match brightness with the other front panel indicators.. If you wish to observe the action of the DCD signal generated by the new circuit on the built in front panel LED, you will have to do the interface a bit differently. First, you will get the negative true CDT signal from pin 1 of JPD. Then insert the LED output signal from the new circuit at either pin 2 of JPD or pin 2 of J5. Remove the jumper currently installed at JPD on the TINY-2 circuit board. If the new circuit is interfaced in this manner, the "RFDCD" signal can no longer be used. This is no great loss, however, as it will also no longer be necessary. (continued in "CIRCUIT_5 EOF Msg # 12776 Type:B Stat:$ To: ALL @OKIPN From: N7CL Date: 28-Aug/2318 Subject: Improved DCD circuit_5 Bulletin ID: 6167_W1FJI Path: AD8I!WB8ICL!N8GTC!KI4UN!KF4NB!WB9TPG!K3VYY!K4EID!K4EID!KE7CZ!W1FJI ******************* FILE 5 of 6 FILES ********************** (C) 31 JUL 88 Eric Gustafson, N7CL 2018 S. Avenida Planeta Tucson, AZ 85710 DPLL Derived Data Carrier Detect (DCD) For Filter Based and Single Chip Modems COMMERCIAL TNC SIGNAL LOCATIONS (continued) Kantronics KAM Interfacing anything to a Kantronics box isn't a job, it's an adventure! Kantronics has an official policy of discouraging anyone from hooking any third party device to their TNCs. This includes external modems of any kind (Never mind that their crystal ball has proven cloudy at best in the past when trying to predict what modems might be popular or necessary in the future). This policy was enunciated to me by persons in their technical support department in two separate telephone conversations. So it was not surprising to find that they didn't provide a modem disconnect header in the KAM. What I did find a little surprising, however, was the fact that they also refuse to provide an individual owner any assistance with signal locations. They don't say they don't know, they say they WON'T help you! If you want, for instance, to interface a JAS-1 (FO-12) style BPSK modem to your Kantronics TNC, you are on your own as far as Kantronics is concerned. Potential Kantronics buyers who are interested in working digital modes through this and the upcoming MICROSAT packet store and forward satellites should take note. If Kantronics thought that one day they might possibly make a radio, you would have to use the optional Kantronics built-in radio in all their TNCs. Thank goodness they don't also make computers... It turns out that the necessary signals ARE available (for 1200 baud at least) in the KAM. It is indeed possible to interface either the 1200 baud BPSK / MANCHESTER FM modem required for the JAS-1 bird or this DCD circuit (or both) to the KAM. At this time it is unclear whether the required clock signal is available for the DCD circuit to operate at 300 baud on this TNC. Even if it is, it would be more trouble than it is worth to interface as it would either require two separate DCD circuits or a switching arrangement to allow the use of one for both modems. Since it is unlikely that the filter / slicer modem used in this box is a stellar performer when working with small shift to baud rate ratio signals of the type used for HF packet, maybe we should only really concern ourselves with 1200 baud operation anyhow. It is worth noting that for wider shift to baud rate ratio signals like RTTY, ASCII, and AMTOR the filter / slicer type demodulator performance is perfectly adequate. When the shift to baud rate ratio is greater than 1, as with these modes, most of the transmitted signal energy is concentrated around and very close to the two tone frequencies. When this is the case, the filter / slicer is the preferred method of demodulation. As these modes do not operate in a Carrier Sense Multiple Access (CSMA) environment like packet requires, the built in CD function is adequate for these modes as well. For 1200 baud operation then, the signal location points of interest in the KAM are as follows: The Receive Data (RXD) signal is obtained from pin 8 of the TCM3105 modem chip. The Kantronics schematic shows what appear to be some numbered pads (17 and 18) on this lead to the processor. If you can locate these points on the circuit board, it may be easier to obtain the signal from one of these points. The X16 baud clock signal is obtained from pin 2 of the TCM3105. The POSITIVE TRUE Carrier Detect (CDT) signal from the modem is obtained from pin 3 of the TCM3105. This line from the modem to the CPU is labeled with 2 numbered pads (7 and 8). The connection between these 2 locations should be broken. JMP2 on the new DCD circuit will be used. The DCD output from the new circuit is injected at pin 21 of the 63B03 CPU. The front panel LED which normally indicates the CDT signal activity will show the action of the new DCD circuit. (continued in "CIRCUIT_6 EOF Msg # 12777 Type:B Stat:$ To: ALL @OKIPN From: N7CL Date: 28-Aug/2320 Subject: Improved DCD circuit_6 Bulletin ID: 6168_W1FJI Path: AD8I!WB8ICL!KI4UN!KF4NB!WB9TPG!K3VYY!K4EID!K4EID!KE7CZ!W1FJI ******************* FILE 6 of 6 FILES ********************** (C) 31 JUL 88 Eric Gustafson, N7CL 2018 S. Avenida Planeta Tucson, AZ 85710 DPLL Derived Data Carrier Detect (DCD) For Filter Based and Single Chip Modems Figure 1 ASCII Representation of DCD Circuit Schematic //=================7 wire BUS * =================\\ +5 V>------------+-------------+--+--+--+--+-----------------+------+--|(-+ G supply | | | | | | | | 10 MF +----------------+ +----------------------+ +--------+ | //-| 3 20 2|---|0 28 27 26 23 1 11|-\\ | 14 | | //-| 4 5|---|9 12|-\\ | U3 | | //-| 7 U1 6|---|8 U2 13|-\\ | 74HC14 | | //-| 8 9|---|7 15|-\\ | 7 | | //-|13 74HC374 12|---|6 27C64 16|-\\ +--------+ | //-|14 15|---|5 17|-\\ G | //-|17 16|---|4 18|-\\ | DATA>-|18 11 10 1 19|---|3 25 24 22 21 20 2 19|-/\/\/\/---+ | from +----------------+ +----------------------+ 4.7K | | modem chip | | | | | | | | | | | | +--+ +--+--+--+--+--+ | | CLK >-------+ | | | | input from | | | Q1 | Q2 X16 or X32 baud G G | E-+-E clock source | / 2N \ +-B 3906 B-+ GND >---G \ (2) / | C-+-C | 1N4148 | | DCD <--------+ +--->|-----+ +---------------+ | out U3D | U3C U3B | | U3A | | ___ 8 /| | 6 /| 4/| | 220K | 2 /| | 470K | LED <--o< +--+--o< |-+---o< |---+-/\/\/\/--+--o< |--+--/\/\/\/--G | out \| \| | \| | \| | | ___ | | 0.47uF | 0.1uF | DCD <----------------+ +---|(----G +--|(-----G | out | 100K | +--/\/\/\/--G | U3F | __ |\ 10K | CD >--+-->o------+------------------< CD <---------+-----/\/\/\/---------+ from |/ | JMP1 | modem | (AMD7910) | chip | |\ | 1110 +---|-->o----------< CD <---------+ |/ JMP2 U3E (TCM3105) NOTE: Only JMP1 OR JMP2 installed NOT both! * BUS wire connections are in order vertically. Top pin goes to top pin bottom pin goes to bottom pin etc. Note: This schematic was restored by AD8I, 31 August 1988. It represents the best attempt to provide full data on the DPLL DCD. Anyone contemplating construction of this circuit should contact N7CL or the TAPR office for an official copy of the circuit and any modification data. ... AD8I